Cisco Cloud Application Policy Infrastructure Controller Release Notes, Release 25.0(2)

January 18, 2022

Release 25.0(2f) became available. In the Resolved Issues section, added bug CSCwa50116.

December 17, 2021

Release 25.0(2e) became available.

Support for importing existing AWS cloud VPCs into Cisco Cloud APIC in unmanaged mode

This release provides support for importing existing brownfield AWS cloud VPCs (VPCs that were not configured through Cisco Cloud APIC) into Cisco Cloud APIC.

For more information, see Importing Existing Brownfield AWS Cloud VPCs Into Cisco Cloud APIC.

Support for external networking using AWS transit gateways

Beginning with release 25.0(2), several external networking enhancements are available for the AWS transit gateway feature:

Cloud APIC on AWS supports network level, account level, and custom routing tables

Support for external branch connectivity over AWS transit gateway with IPsec

For more information, see Increasing Bandwidth Between VPCs by Using AWS Transit Gateway or AWS Transit Gateway Connect.

Support for site-external EPGs using Azure VPN gateways

Support is available for providing connectivity between a Cloud APIC-managed cloud site and a non-ACI remote site using VPN gateway.

For more information, see Cisco Cloud APIC for Azure User Guide, Release 25.0(x).

Support for all availability zones in an AWS region for user tenants

Support is now provided for multiple (greater than two) availability zones in AWS for Cisco Cloud APIC.

For more information, see Cisco Cloud APIC for AWS User Guide, Release 25.0(x).

Support for configuring routing and security policies independently in Azure and AWS

Beginning with release 25.0(2), the following updates are available for the routing policies:

Support for route maps-based route leaking between a pair of internal VRFs

Support for a global internal VRF route leak policy, which allows you to choose whether you want to use contract-based routing or route map-based routing between a pair of internal VRFs.

For more information, see:

·       Cisco Cloud APIC for AWS User Guide, Release 25.0(x)

·       Cisco Cloud APIC for Azure User Guide, Release 25.0(x)

CSR IPsec tunnels can now use any of the three available data interfaces for external branch connectivity

Prior to release 25.0(2), all the tunnels to external networks are originated from one specific interface on the CSR router (the GigabitEthernet3 interface, or cloudHostIfp-2). Beginning with release 25.0(2), support is now extended where tunnels to the same destination can be formed from the GigabitEthernet2, GigabitEthernet3, and GigabitEthernet4 interfaces. This is supported for tunnels with IKEv2 configurations only.

For more information, see:

·       Cisco Cloud APIC for AWS User Guide, Release 25.0(x)

·       Cisco Cloud APIC for Azure User Guide, Release 25.0(x)

Support for VM scale sets for Azure NLB backend pools

Beginning with release 25.0(2), support is added for Azure virtual machine scale sets as backend targets for load balancers.

For more information, see Cisco Cloud APIC for Azure User Guide, Release 25.0(x).

Support for increased number of cloud regions for workload deployment

Prior to release 25.0(2), you can have a maximum of four regions per site. Beginning with release 25.0(2), you can have a maximum of sixteen regions per site.

For more information, see:

·       Cisco Cloud APIC for AWS User Guide, Release 25.0(x)

·       Cisco Cloud APIC for Azure User Guide, Release 25.0(x)

CSCvy07759

CSR upgrade banner is not updated after the upgrade is complete.

25.0(1c) and later

CSCvz31331

APIC REST APIs allows to create a cloudEPg that refers to an external VRF in the infra tenant.

This is disabled when configuring through the GUI and should be blocked in the backend as well.

25.0(1c) and later

CSCvz62225

When you scale up the number of CSRs or routers per region, some of the configurations may be missing on the newly created CSR. This issue happens randomly on the newly created CSRs, in this case tunnels or BGP sessions on the new CSRs may be down due to missing configuration.

25.0(1c) and later

CSCvz66172

Unable to get public IP addresses assigned to non Gig1 Interfaces of CSR. Gig1 Interface gets a public IP addresses.

25.0(1c) and later

CSCvz47166

When delete followed by add operations of tenant or other resources (such as VPCs and contracts) are done within a short span of time, it is possible that the resource deployment in Google Cloud may get out of sync with the configuration on Cisco Cloud APIC. The likelihood of this happening is directly proportionate to the scale of configuration and how quickly the operations are done.

We may see resources either not created or not deleted on Google Cloud to match the user configuration on Cisco Cloud APIC.

25.0(1c) and later

CSCvy97972

Cisco Cloud APIC in this release limits the number of regions where we can deploy the hubnetwork in order to establish external connectivity. When you attempt to deploy/configure hubnetwork in more than four regions, the configuration will be rejected with the following error:

Invalid Configuration CT_INTNETWORK_REGION_MAXIMUM: At present, there can be at most 4 cloudRegionName in cloudtemplateIntNetwork uni/tn-infra/infranetwork-default/intnetwork-default; current count = <total-hubnetwork-regions-attempted>

25.0(1c) and later

CSCvz17160

Customers are restricted to shorter key value pairs than they need to be.

25.0(1c) and later

CSCvz21771

VPN tunnels may not come up when the Cisco Cloud APIC configuration is posted via XML interface. This problem won't be seen/encountered when we use the Cisco Cloud APIC UI.

25.0(1c) and later

CSCvz26752

The Cisco Cloud APIC UI may display empty entries when routes are leaked to or from non-existing VRFs. For example, this is seen when a VRF is deleted and another VRF leaked one or more routes to that deleted VRF. The UI may indicate an empty value under the VRF to which the routes are leaked.

25.0(1c) and later

CSCvz41009

When delete followed by add operations of tenant or other resources (such as VPCs and contracts) are done within a short span of time, it is possible that the resource deployment in Google Cloud may get out of sync with the configuration on Cisco Cloud APIC. The likelihood of this happening is directly proportionate to the scale of configuration and how quickly the operations are done.

We may see resources either not created or not deleted on Google Cloud to match the user configuration on Cisco Cloud APIC.

25.0(1c) and later

CSCvz43324

The Cloud APIC REST APIs allow you to create a cloud EPG that refers to an External VRF in the infra tenant.

This is disabled on the UI and should be disabled through the REST API as well.

25.0(1c) and later

CSCvz46464

There may be some traffic loss encountered when Cisco Cloud APIC is rebooted and a new configuration is imported. The new configuration takes time to deploy as we achieve eventual consistency.

25.0(1c) and later

CSCvz47232

When we delete any configuration, it takes time to reach eventual consistency and clean up all the resources on the cloud. This may not happen when Cisco Cloud APIC is rebooted while the delete/cleanup operation is underway. A few resources may remain on the cloud if Cisco Cloud APIC is rebooted while the cleanup is in progress.

25.0(1c) and later

CSCvy89617

Cloud routers may not get created if external network objects are not configured. External network configuration is required for configuring cloud routers.

25.0(1c) and later

CSCvy94328

verlay-1 VRF in tenant "infra" shows up as one of the VRFs with which an external network is associated. However, Cisco Cloud APIC does not allow overlay-1 VRF to be associated with any external network. In this release, overlay-1 VRF shows up alongside other VRFs to be associated with an external network. It has no functional impact, but it gives an incorrect impression that we have a external network associated with overlay-1 VRF. This external network name is set to default and there are no other objects/MOs for this external network configured.

25.0(1c) and later

CSCvz38067

Incorrect DNS server is configured on Cisco Cloud APIC with Google Cloud. Though this is not directly used when deploying Cisco Cloud APIC with Google Cloud, an incorrect IP address is configured.

25.0(1c) and later

CSCvz11574

The cloud VRF egress route table is missing the route for 0.0.0.0/0 via the Internet Gateway (IGW), which leads to issues with ssh for VMs in the cloud VRF.

25.0(1c) and later

CSCvz52773

When performing a Cisco Cloud APIC upgrade (but not also performing a CSR upgrade), before the upgrade is finished and when the Cisco Cloud APIC is reconciling the CSR configurations, if you delete certain configurations and add the same configurations back (for example, if you delete a VRF and add the VRF back), a traffic drop may happen. Eventually it should recover.

25.0(1c) and later

CSCvz39389

After a clean reboot of Cisco Cloud APIC and an import of the configuration, a CSR might take around 45 minutes to re-establish the datapath readiness.

25.0(1c) and later

CSCvz40326

Routes to and from overlay-2 VRF may not be configured in the cloud deployment.

25.0(1c) and later

CSCvo30542

TACACS monitoring of the destination group is not supported through the GUI.

25.0(1c) and later

CSCvu64277

Stats seen on Cisco Cloud APIC are sometimes not in sync with Azure stats.

25.0(1c) and later

CSCvu66521

In the "Cloud Resources" section of the GUI, the names displayed in the "Name" column are not the same as the name of resources on the cloud. These are showing the Cloud APIC object names.

25.0(1c) and later

CSCvu72354

Adding an EPG endpoint selector fails with an error message saying the selector is already attached.

25.0(1c) and later

CSCvu78074

Route nextHop is not set to the redirect service node specified in the service graph.

25.0(1c) and later

CSCvv32664

When the CSR bandwidth needs to be increased, the user needs to undeploy all the CSRs in all the regions and redeploy with the desired bandwidth, which can cause traffic loss.

25.0(1c) and later

CSCvx16601

When the "AllowAll" flag is enabled on a service device such as a native load balancer or on the logical interface of a third party device, it is possible that to see some specific rules apart form a rule that allows all traffic from any source to any destination.

25.0(1c) and later

CSCvy06610

The eventmgr crashes when handling a fault triggered by a new cloud account.

25.0(1c) and later

CSCwa03277

When the brownfield VPC is imported into Cisco Cloud APIC, you need to take care of the creation and management of the route table and route table entries, security group rules, and transit gateway VPC attachment.

After creating the transit gateway VPC attachment with the infra transit gateway for the brownfield VPC in the AWS console, the corresponding  cloudCtxPeerOper for the brownfield cloud context profile will move from Failed state to Configured state.

After that, if the created transit gateway attachment for the brownfield VPC is deleted in the AWS console, cloudCtxPeerOper is not moving back to Failed state.

25.0(2e) and later

CSCwa40843

Stale IPsec configuration remains when an IKEv1 IPsec tunnel is deleted.

25.0(2e) and later

CSCwa07078

When the brownfield VPC is imported into Cisco Cloud APIC via REST API POST, a new cloud context profile is created. Under this cloud context profile, cloudRsCtxToAccessPolicy is created, which is in relation to read only access policy. One or more cloudCIDRs and cloudBrownfield with cloudIDMapping, which holds the VPC ID, is posted to Cloud APIC.

cloudIDMapping, which contains the VPC ID, points to the VPC present in the cloud. If the VPC ID is non-existent or if there is any difference between the cloudCIDRs posted vs the CIDRs present in the VPC, cloudCtxOper and cloudCidrOper moves to a Failed state.

But because of the delegate's distinguished name, the imported unmanaged VPC shows healthy.

25.0(2e) and later

 CSCwa10752

When CSRs are deployed in non-home regions, and no CSR is deployed in the home region (where Cloud APIC is deployed) in Azure, faults are seen in the Cloud APIC where the ssh connectivity to the CSR is down.

25.0(2e) and later

CSCwa29007

In the TGW l3out configuration, modifying the IPsec tunnel pre-shared key is not supported. UI does not allow the modification of the IPsec tunnel pre-shared key, as this field is grayed out.

API accepts this modification of the IPsec tunnel pre-shared key, but it's not updated on the AWS.

25.0(2e) and later

 CSCwa08564

UI dashboard shows the wrong status for inter-region connectivity.

25.0(2e) and later

CSCwa40705

When an IKEv1 tunnel is configured to a destination while another tunnel already exists to the same destination but with a different source interface, this tunnel will remain with the protocol shown as down.

25.0(2e) and later

CSCwa28888

This issue is hit in some cases where Cloud APIC is unable to deploy infra configuration, such as creating cloud routers in overlay-1 VPC. This is sometimes seen in new deployments, but not in the case of upgrade scenario. Cloud routers and other configurations do not get deployed in Google Cloud.

25.0(2e) and later

CSCwa44822

When the tunnels are created from source interface Gig 2 or 4, we put allow all for security group. In future the allow all rule needs to be removed and provide explicit IP needs to be allowed.

25.0(2e) and later

CSCwa36940

Transit gateway external connectivity is not getting deployed in regions where cloud context profiles are deployed.

25.0(2e) and later

CSCwa54001

When a quick delete and add of inter-site connectivity happens in such a way that the tunnels on the cloud are unchanged, the status on the UI may show tunnels as down even when the tunnels are up on the GCP cloud and the Azure CSR.

This issue might also happen during an upgrade and/or connector restart or a crash.

25.0(2e) and later

CSCwa43845

This issue occurs if there is a misconfiguration done where the local subnet was provided under routes leaked from external VRF to internal VRF.

This is an Azure Cloud APIC only issue, since AWS does not allow programming routes that overlap with VPC's CIDR.

25.0(2e) and later

CSCwa48929

If Cloud APIC is rebooted, in a rare case an expected rule may not be programmed on the cloud due to a timing mismatch in the reconcile and programming workflow.

25.0(2e) and later

CSCwa49263

In VPC route table, the route table entry for a destination CIDR pointing to transit gateway is sometimes missing when a quick delete and add of tenant or contract is done or when we move from transit gateway connect to legacy transit gateway solution. This happens only with legacy transit gateway solution in either of the cases.

This is a timing issue with the legacy transit gateway solution, where we create two transit gateways per hub network in a region. This can happen either if we move to legacy transit gateway solution or if legacy transit gateways are coming up for the first time.

These conditions result in deleting the route table entry for a given destination CIDR and adding back the same entry at the same time. Due to an issue with the AWS API which returns a deleted route table entry as a non deleted entry, Cloud APIC deletes the wrong entry.

25.0(2e) and later

CSCwa49534

Consider 3 regions (A, B, C), where both A and B have CSR in their region.

Transit gateway cross-region comes in picture when there are no local CSRs in a region(region 'A'). The traffic of region "A" will hit the local transit gateway. On this transit gateway's user VPC Route Table, there will be a 0/0 entry which will have next-hop pointing to the nearest regions (region "B") transit gateway peering attachment.

When the CSRs of region "B" go down, the 0/0 route in the region "A" transit gateway route table should now point to take the next hop to region "C" transit gateway peering.

But in this bug, this flip does not happen at all times, which leads to traffic loss.

25.0(2e) and later

CSCwa26716

An external endpoint group of type non internet cannot leak all routes(or public internet IPs) to a cloud endpoint group. This results in creating a static route in the Vnet route table to point to CSR NLB for that destination CIDR. If it's leak-all, we configure 0.0.0.0/0 to point to CSR NLB.

In case of Azure Cloud APIC we do not create a static route to internet, Azure does this implicitly by default. Cloud APIC only programs the rules. When a user creates a contract or route leak, Cloud APIC programs a static route to CSR NLB. This overrides the Azure implicit default route to internet if the CIDR overlaps with the pubic internet IPs, since the leak-all creates a 0.0.0.0/0 route to point to CSR NLB.

This is an invalid configuration unless the user intended it, i,e the user intended to login to the VM through it's private IP through CSR. If not the user has to leak specific routes to the cloud endpoint group.

25.0(2e) and later

CSCvz87367

Even though the preferred DNS server is taking into effect in Cloud APIC, the DNS server which was configured first continues to be the preferred DNS server.

25.0(2e) and later

CSCwa45047

This is not a functional issue. There will be no fault shown in the Cloud APIC UI if the border gateway protocol sessions of the transit gateway external connectivity are down.

25.0(2e) and later

CSCwa50116

This bug has been filed to evaluate the Cisco Cloud APIC against the vulnerability in the Apache Log4j Java library disclosed on December 9th, 2021.

Cisco has reviewed this product and concluded that it contains a vulnerable version of Apache Log4j and is affected by the following vulnerability:

CVE-2021-44228 - Apache Log4j2 JNDI features do not protect against attacker controlled LDAP and other JNDI related endpoints

This advisory is available at the following link:

https://tools.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-apache-log4j-qRuKNEbd

25.0(1c) and 25.0(2e)

CSCwa18353

The route between the internal and external VRF is not programmed on the CSR. It is expected to be configured on CSR as a part of the leakTo subnet configuration.

25.0(2f)

CSCwa50116

This bug has been filed to evaluate the Cisco Cloud APIC against the vulnerability in the Apache Log4j Java library disclosed on December 9th, 2021.

Cisco has reviewed this product and concluded that it contains a vulnerable version of Apache Log4j and is affected by the following vulnerability:

CVE-2021-44228 - Apache Log4j2 JNDI features do not protect against attacker controlled LDAP and other JNDI related endpoints

This advisory is available at the following link:

https://tools.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-apache-log4j-qRuKNEbd

25.0(2f)

CSCvy42684

Importing a configuration into Cloud APIC 5.2 displays the following error: maximum buffer length exceeded.

25.0(2e)

CSCvz49747

When delete followed by add operations of tenant or other resources (such as VPCs and contracts) are done within a short span of time, it is possible that the resource deployment in Google Cloud may get out of sync with the configuration on Cisco Cloud APIC. The likelihood of this happening is directly proportionate to the scale of configuration and how quickly the operations are done.

We may see resources either not created or not deleted on Google Cloud to match the user configuration on Cisco Cloud APIC.

25.0(1c) and later

CSCvy77233

Routes for subnets that are not yet configured in Google Cloud may become visible on an external device. When you configure routes to be advertised to an external device, but don't actually configure subnets in the cloud that you intend to advertise the routes for, those routes are still advertised.

Remote router may see routes that are advertised even when the subnets are not yet configured.

The traffic will get dropped because the subnets are not actually configured.

25.0(1c) and later

CSCvx98260

When delete followed by add operations of tenant or other resources (such as VPCs and contracts) are done within a short span of time, it is possible that the resource deployment in Google Cloud may get out of sync with the configuration on Cisco Cloud APIC. The likelihood of this happening is directly proportionate to the scale of configuration and how quickly the operations are done.

We may see resources either not created or not deleted on Google Cloud to match the user configuration on Cisco Cloud APIC.

25.0(1c) and later

CSCvo06626

When a cloudExtEpg matches on a 0/0 network and has a bi-directional contract with two cloud EPGs, such as cloudEpg1 and CloudEpg2, this can result in inadvertent communication between endpoints in cloudEpg1 and cloudEpg2 without a contract between the two EPGs themselves.

25.0(1c) and later

CSCvo55112

Logs are lost upon stopping the Cloud APIC instance.

25.0(1c) and later

CSCvo95998

There is traffic loss after a Cloud APIC upgrade. Traffic will eventually converge, but this could take a few minutes.

25.0(1c) and later

CSCvq11780

Creating VPN connections fail with the "invalidCidr" error in AWS or the "More than one connection having the same BGP setting is not allowed" error in Azure.

25.0(1c) and later

CSCvq76039

When a fault is raised in the Cloud APIC, the fault message will be truncated and will not include the entire cloud message description.

25.0(1c) and later

CSCvr01341

REST API access to the Cloud APIC becomes delayed after deleting a tenant with scaled EPGs and endpoints. The client needs to retry after receiving the error.

25.0(1c) and later

CSCvu05329

The Ctx Oper managed object is not deleted after the attachment is deleted.

25.0(1c) and later

CSCvu81355

Traffic gets dropped after downgrading to the 5.0(1) release. Cloud Services Router has incompatible configurations due to an issue with reading configurations using SSH.

25.0(1c) and later

CSCvu88006

On the Dashboard, fewer VNet peerings are shown than expected.

25.0(1c) and later

CSCvv81647

When an invalid Cloud Services Router license token is configured after initially configuring a valid token, the Cloud Services Router fails the license registration and keeps using the old valid token. This failure can only be found from the CSR event log.

25.0(1c) and later

CSCvw05821

Redirection and UDR does not take effect when traffic coming through an express route and destined to a service end point is redirected to a native load balancer or firewall.

25.0(1c) and later

CSCvw07392

Inter-site VxLAN traffic drops for a given VRF table when it is deleted and re-added. Packet capture on the CSR shows "Incomplete Adjacency" as follows:

Punt     1    Count       Code  Cause    1           10    Incomplete adjacency  <<<<<<<

Drop     1    Count       Code  Cause    1           94    Ipv4NoAdj

25.0(1c) and later

CSCvw07781

There is complete traffic loss for 180 seconds.

25.0(1c) and later

CSCvw24376

Inter region traffic is black-holed after the delete trigger for contracts/filter. It was observed that the TGW entry pointing to the remote region TGW is missing for the destination routes. On further debugging it was found that post delete trigger as part of re-add flow, when a describe call is sent to AWS got a reply with the state of this entry as "active" because of which a new create request is not being sent.

25.0(1c) and later

CSCvw39814

Infra VPC subnet route table entry for 0.0.0.0/0 route with TGW attachment as nh, is left as a stale entry upon being undeployed. There is no functional impact. Upon being redeployed, this entry is updated with the correct TGW attachment ID as nh.

25.0(1c) and later

CSCvw40737

SSH to a virtual machine's public IP address fails, despite the NSG allowing the traffic inbound. SSH to the private IP address of the virtual machine from within the VNet works.

25.0(1c) and later

CSCvw40818

After upgrading Cloud APIC, the Cloud Services Routers will be upgraded in two batches. The even set of CSRs are triggered for upgrade first. After their upgrade is complete and all of the even CSRs are datapathReady, only then the odd set of CSRs will be triggered for upgrade. When even one of the upgrade of the even CSRs fail and they don't become datapathReady, the odd set of CSRs will not be triggered for upgrade. This is the behavior followed to avoid any traffic loss.

25.0(1c) and later

CSCvw48190

When Cloud APIC is restart, the VPN connection from a tenant's VNets will get deleted and re-created, one by one. This can be seen in the Azure activity logs. It should not impact traffic, as all connections are not deleted at the same time.

25.0(1c) and later

CSCvw49898

When the downgrading from the 5.2(1) release to the 5.0(2) release, traffic loss is expected until all of the CSRs are downgraded back to the 17.1 release. The traffic loss occurs because when the CSRs are getting downgraded to the 17.1 release, the CSR NIC1s will be in the backendPools and traffic from the spokes will still be forwarded to the native load balancer. The traffic gets blackholed until the CSRs get fully programmed with all the configurations in the 17.1 release.

25.0(1c) and later

CSCvw50918

Upon downgrading Cloud APIC, VPN connections between Cloud APIC and the cloud (AWS/Azure VPN gateway) will be deleted and re-created, causing traffic loss. Traffic loss is based on how quickly the VPN connections are deleted and re-created in AWS due to AWS throttling.

25.0(1c) and later

CSCvw51544

A user who is assigned a large number of security domains may not be able to create other Cisco ACI policies.

25.0(1c) and later

CSCvw55088

A user who is assigned a large number of security domains may not be able to create other Cisco ACI policies.

25.0(1c) and later

CSCvx91010

When TGW Connect is disabled, traffic loss is observed for about 8 minutes.

25.0(1c) and later

CSCvy10936

Downgrading Cisco Cloud APIC from release 5.2(1) to 5.1(2) may cause CSRs to not be downgraded. The CSR release for 5.2(1) is 17.3.2, and the CSR version for release 5.1(2) is 17.3.1. After the Cisco Cloud APIC downgrade, the CSR version should be downgraded to 17.3.1,  but it will not happen due to this bug.

25.0(1c) and later

CSCvy12722

Loss of traffic between a cloud and Cisco ACI On-Premises deployment.

25.0(1c) and later

CSCvy13369

After upgrading AWS, infra vPC peering does not get deleted.

25.0(1c) and later

CSCvy19286

There is traffic loss after downgrading from 5.2(1) to 5.1(2).

25.0(1c) and later

CSCvy28890

There is a loss in SSH connectivity to the Cisco Cloud APIC across reboots. But, after a few minutes, the connection should come back and users will be able to SSH in to the Cisco Cloud APIC again.

25.0(1c) and later

CSCvy28896

There is an increase in the connector's memory utilization. All of the CSR workflows rerunning might happen even after the setup is in the steady state.

25.0(1c) and later

CSCvy30314

After upgrading the Cisco Cloud APIC, on the TGW route tables, the default route (0.0.0.0/0) does not point to infra VPC attachment or is missing. In this case, traffic intended to get forwarded to the CSR will be dropped or forwarded to an invalid next-hop.

25.0(1c) and later

CSCvy33435

There is intersite traffic loss when TGW Connect is enabled.

25.0(1c) and later

CSCvy34180

Cloud Intersite traffic is dropped due to the CSR in the cloud site not advertising the EVPN routes.

25.0(1c) and later

CSCvy45517

The Cisco Cloud APIC GUI shows the total allowed count for CtxProfile, VRF (fvCtx), EPGs, and contracts. These numbers have been validated only for Azure-based deployments. For AWS deployments, the numbers supported are much lower.

25.0(1c) and later

CSCvz20282

An upgrade to or downgrade from the Cloud APIC 5.2(1g) release to any release while using "Ignore Compatibility Check: no" will fail. The following fault is raised: "The upgrade has an upgrade status of Failed Due to Incompatible Desired Version."

25.0(1c) and later